Past and present research

My most significant contributions to understanding the Earth's interior have come through a combination of observational and interpretational work on seismic data. This includes the introduction of innovative methods and techniques, and new data. I have been interested in improving spatial sampling by seismic waves of various parts of the Earth's interior, such as the inner core, the lowermost mantle and the lithosphere. During my graduate student days at the University of California at Berkeley, I assembled and analyzed a high-quality global dataset of differential travel times of the core- and deep mantle-sensitive seismic waves. In 1998, with colleagues L. Breger and B. Romanowicz, we observed distinct trends in PKP differential travel time residuals (when the latter were plotted as a function of the angle between Earth's rotation axis and PKP leg in the inner core), by distinguishing among several different geographical samplings of the corresponding paths in the mantle. We showed that most of the trends observed in PKP differential travel time residuals stem from heterogeneities in the lowermost mantle (2000). In particular, we were able to assess how much of the core-sensitive PKP travel-time data can be explained by mantle structure alone. Using PKP and PcP-P differential travel time data, I developed a 3-D compressional velocity model of the lowermost mantle (2002), and outlined its relevance for understanding anisotropic structure of the Earth's core.
Faced with a lack of volumetric sampling of the deepest portions of Earth by PKP waves, I have been particularly interested in identifying various multiple core seismic phases such as PKPPKP, PnKP, etc. that are normally very difficult to observe, even on broadband records. In 2004, I carried out a systematic search over the existing waveform records in order to observe very podal PKPPKP waves (at epicentral distances less than 10 degrees) as such ray paths would sample the inner core in a unique way. This search resulted in previously unobserved (although theoretically predicted) podal PKPPKP waves, on both short-period and broadband records, on individual as well as on array stations. The results indicated that most PKPPKP arrivals are not anomalously advanced with respect to the standard 1D model of Earth (ak135). In addition, I observed high-energy precursors to the main PKPPKP arrivals. With colleagues M. Flanagan and V. Cormier, I demonstrated that these precursors are a result of a back-scattered energy from a horizontally connected small-scale heterogeneity concentrated in the uppermost 150-220 km of the mantle (2006).
I directed my efforts towards finding new ways to analyze existing low-quality PKP data. With colleagues R. Garcia and S. Chevrot, we applied a novel non-linear inversion approach to measure core-sensitive seismic phases from available seismograms (2006). This allowed us to consider shallow, as well as mid-ocean ridge events that previously had to be discarded due to low signal to noise ratio, and to collect and add new geometries to samplings of the lowermost mantle and the core. In Australia, there is a great potential to extract valuable information about Earth's deep interior using the seismic data that accumulated over the years of carefully planned seismic deployments across the continent.
Over the last several years, I studied the Earth's core density contrast using new data from earthquakes and nuclear explosions, and on core structure and anisotropy using new observations from Antarctica deployed by A. Reading. We found from these travel time data that the inner core is weakly anisotropic on average (2010), and confirmed that the South Sandwich Islands earthquakes observed in Alaska probably represent a significant anomaly that led to a widely accepted model of strongly anisotropic inner core. I proposed a model according to which the inner core is a conglomerate of anisotropic domains (2010), which challenged old paradigms. Further studies with colleagues B. Kennett and V. Cormier (2009, 2010) revealed that the inner core boundary is likely laterally variable and not a simple boundary. I am currently collaborating with S. Tanaka on modelling the inner core boundary from unique observations of PKiKP and PcP waves.
Most recently, by observing new earthquake doublets and analysing data within the Bayesian inference framework with students and colleagues at the ANU, I discovered that the inner core changes its rotation rate in time in the reference frame of the Earth's mantle, by rotating either faster or slower than the rest of the mantle. This has been seen as a breakthrough in reconciling old discrepancies between the body wave and normal mode observations, and an invaluable observational constraint of how the Earth's geodynamo works.

I have been working on developing reliable methods to study lithospheric structure using receiver functions and surface wave dispersion data. These methods have been applied with colleagues at Lawrence Livermore National Laboratory to study lithospheric structure of the Middle East (2006), and with a number of colleagues in studies of lithospheric structure of Australia (2012, 2013), China (2010, 2011), Croatia with the Adriatic Sea (2011) and the island La Reunion (2013). With T. Bodin and M. Sambridge, we have recently presented a new method (2012), where the number of free parameters in the inversion (e.g. the number of Earth layers) becomes a free parameter itself, and the data noise is also treated as a free parameter. This method is known as a transdimensional hierarchical Bayesian inversion of receiver functions.

My seismic moment tensor solutions are part of the UC Berkeley Catalogue for the period from 1997 to 2001. With D. Dreger (2000), he showed the existence of strong volumetric components in volcanic earthquakes from the Long Valley Caldera, which indicated a direct relationship between the seismicity and hydrothermal or magmatic processes. With colleagues D. Dreger, G. Foulger and B. Julian, he demonstrated that volumetric components were not present (contrary to expectations) in the moment tensor of the anomalous Bardarbunga Volcano, Iceland, earthquake (2009). The volumetric components are not likely to be masked by the presence of the system of two magma chambers or one magma chamber and an opening crack, therefore a rupture along a conical surface fault is a plausible candidate for this earthquake with strong vertically oriented CLVD component. Such a rupture on a boundary ring fault to explain anomalous seismic radiation observed on HOTSPOT recordings was successfully modeled (2004) using a finite difference method developed by S. Larsen. These experiments, along with the results of a finite source probabilistic study with A. Fichtner indicate that a super shear rupture took place and resulted in a collapse of the magma chamber (2011).

Future research concepts

As an observational seismologist at RSES, I have active interest in various aspects of seismology. For example, I am genuinely interested in making contributions toward ongoing efforts at RSES for achieving an unprecedented coverage of a continental lithosphere by seismic deployments and individual stations. I am engaged in a further proliferation of seismic instruments and data in Australia, which will establish a solid basis for inventing new approaches in studying Earth's structure as well as seismic sources. In the following, three main focuses in these studies are described.

Deep Earth structure

Inadequate spatial sampling of the central inner core by PKP waves in all directions makes further advances in understanding anisotropic properties (especially anisotropy's radial dependence and a hemispherical pattern observed for inner core velocity and attenuation) very difficult. One of the reasons for this incomplete sampling lies in the fact that, in order to pass through the central regions of the inner core, PKP waves must be nearly antipodal. With the spatial distribution of large earthquakes and current configuration of the seismographic stations worldwide, this is difficult to achieve, except for the paths nearly parallel to the equatorial plane. I have been working toward improving the spatial sampling of the core and the lowermost mantle by seismic body waves, and using these data to better understand fine details of deep Earth structure (e.g., heterogeneous and anisotropic structure of the inner core, the depth and sharpness of discontinuities). I believe that three major approaches could be pursued to achieve the above-mentioned objective, and they are:

1) Observation and analysis of seismic phases with more complex geometry, such as PKPPKP or PnKP, which must be employed as a necessary supplement to PKP measurements when using seismic travel times to study deep Earth structure (because of their unique sampling of the core that cannot be achieved by PKP waves only);

2) Installation of seismic stations in remote continental areas without previous access, at extreme geographic latitudes and oceanic islands, in order to increase the spatial sampling of the inner core and lowermost mantle by body waves. The Australian seismic deployments across the continent and in Antarctica greatly help in achieving this objective; I am currently managing the operation of the Warramunga Array in Northern Territory, and the installation of two arrays in remote parts of Western Australia and Queensland. Multiple arrays will be used to probe heterogeneity of the deep Earth and help improve lateral resolution of Earth's internal structure.

3) Development and application of new techniques (e.g. array signal processing techniques, or travel time measurements by non-linear inversion instead of direct time picking), which will allow us to use core-sensitive seismic phases and other data that were previously discarded.

In summary, there is a great potential to extract valuable information about Earth's deep interior by using existing and new waveform data and by interpreting these data using various computational methods. For example, an improved anisotropy model of the inner core and an improved P-wave tomographic model of the lowermost mantle are some of the priorities.

Lithospheric structure

Tomographic imaging techniques often result in large uncertainties in vertical resolution of crustal and lithospheric layers. Alternative imaging techniques such as receiver functions may be deployed to improve general understanding of lithospheric structure in various regions of the world. There have been numerous studies addressing the importance of combining surface wave dispersion data with the teleseismic receiver functions. This stems from the fact that the receiver functions are mostly sensitive to sharp gradients in elastic properties, while surface wave data contribute to a better understanding of overall absolute velocity. I have been developing a reliable method of joint modeling of these two types of data through a multi-step procedure that includes both linearized and non-linear inversion approaches. The multi-step modeling has been applied to the waveform data collected by the seismic networks of various regions to better constrain their lithospheric structure, including features such as the crustal thickness, upper mantle low velocity zone and transverse isotropy (polarization anisotropy). The improved structural models have been derived using the combination of receiver functions and constraints from travel time tomography as well as surface wave and ambient noise dispersion. The improved methods include the Bayesian transdimensional hierarchical inference framework, and we are currently working to improve the capacity to also account for dipping layers.

Seismic sources

I am keenly interested in observing and studying kinematics of seismic sources, especially of the earthquakes and explosions with anomalous seismic radiation and puzzling focal mechanisms. These include but are not limited to volcanic earthquakes, explosions and impacts. I have been using a full seismic moment tensor representation and different computational methods to reveal statistically significant non-double-couple components and model complex finite sources. New methods include the utilization of transdimensional Bayesian inversion recently introduced to geosciences. The improved structural models of Australia and its surroundings will be used for different aspects of a regional moment tensor analysis. Australia and its surroundings is an exciting environment for such studies, because of a great variety of physical mechanisms responsible for earthquakes.